def __init__(self, full_length_lineage, min_rel_abundance,
min_abs_abundance):
self.full_length_lineage = full_length_lineage
self.min_rel_abundance = min_rel_abundance # float number from 0 to 1
self.min_abs_abundance = min_abs_abundance
self.aa_dict = amino_acid_utils.AminoAcidDict()
self.aa_dict.AddLineage(full_length_lineage)
def _InitEdges(self):
self.aa_dict = amino_acid_utils.AminoAcidDict()
self.aa_dict.AddClonalTree(self.clonal_tree)
self.used_aa = []
self.aa_edges = dict()
self.vertex_indices = set()
self.num_edges = 0
for v in self.clonal_tree.VertexIter():
v_id = self.clonal_tree.GetSequenceByVertex(v).id
v_aa = self.aa_dict.GetAAById(v_id)
v_index = self.aa_dict.GetIndexByAA(v_aa)
self.vertex_indices.add(v_index)
for e in self.clonal_tree.EdgeIter():
self.num_edges += 1
src_id = self.clonal_tree.GetSequenceByVertex(e[0]).id
dst_id = self.clonal_tree.GetSequenceByVertex(e[1]).id
src_aa = self.aa_dict.GetAAById(src_id)
dst_aa = self.aa_dict.GetAAById(dst_id)
src_index = self.aa_dict.GetIndexByAA(src_aa)
dst_index = self.aa_dict.GetIndexByAA(dst_aa)
# self.vertex_indices.add(src_index)
# self.vertex_indices.add(dst_index)
if src_aa == dst_aa:
continue
aa_edge = (src_index, dst_index)
if aa_edge not in self.aa_edges:
self.aa_edges[aa_edge] = 0
self.aa_edges[aa_edge] += 1
print "# amino acid vertices: " + str(len(self.vertex_indices))
def __init__(self, full_length_lineage, min_rel_abundance, min_abs_abundance):
self.full_length_lineage = full_length_lineage
self.min_rel_abundance = min_rel_abundance # float number from 0 to 1
self.min_abs_abundance = min_abs_abundance
self.aa_dict = amino_acid_utils.AminoAcidDict()
self.aa_dict.AddLineage(full_length_lineage)
abundant_aa_seqs = []
for aa in self.aa_dict:
if float(self.aa_dict.GetAAMultiplicity(aa)) / len(full_length_lineage) < self.min_rel_abundance or self.aa_dict.GetAAMultiplicity(aa) < min_abs_abundance:
continue
# print 'adding aa with multiplicity ' + str(self.aa_dict.GetAAMultiplicity(aa))
seq_ids = self.aa_dict.GetIdsByAA(aa)
seqs = [self.full_length_lineage.GetFullLengthSequenceByName(seq_id) for seq_id in seq_ids]
abundant_aa_seqs.extend(seqs)
self.seqs = []
self.seqs.append(abundant_aa_seqs)